feat: default amplihack copilot to autopilot mode#3013
Conversation
|
🤖 Auto-fixed version bump The version in If you need a minor or major version bump instead, please update |
|
Automated review summary: code review found and I fixed the forced-autopilot override bug before opening this PR; security scope is limited to local CLI arg assembly; final implementation is a small launcher conditional plus focused tests. |
Default `amplihack copilot` launches now use `--autopilot --yolo --max-autopilot-continues 100` when no Copilot passthrough args are supplied. Passing args after `--` now acts as a full override of those defaults. Refs #3007 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
99dd2b2 to
8b08507
Compare
Repo Guardian - PassedAll 4 changed files are durable project artifacts — no ephemeral content detected.
No meeting notes, investigation journals, one-off scripts, or other point-in-time content found.
|
|
🤖 Auto-fixed version bump The version in If you need a minor or major version bump instead, please update |
Cover merged PRs from the last 24 hours following the Diátaxis framework: - docs/COPILOT_CLI.md: Add 'Using amplihack copilot' section documenting autopilot defaults (PR #3013: --autopilot --yolo --max-autopilot-continues 100) and override mechanism via '--'. Add Copilot CLI tool-policy compatibility notes for tmux restart alternatives (PR #3012). Bump version to 1.2.0. - docs/howto/settings-hook-configuration.md: Add 'Rust Hook Engine' section documenting AMPLIHACK_HOOK_ENGINE=rust (PR #2999), binary search order, hook mapping table, no-fallback policy, and wrapper script generation (audit fixes from PRs #3003, #3005, #3009). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Summary
amplihack copilotto--autopilot --yolo --max-autopilot-continues 100when no Copilot passthrough args are supplied--as a full override of those defaultsCloses #3007
Step 13: Local Testing Results
Test Environment: feature branch
feat/issue-3007-copilot-autopilot-defaultsviauv run pytestTests Executed:
uv run pytest tests/test_copilot_launcher_integration.py -k 'autopilot_defaults or full_override' -q→ ✅uv run pytest tests/test_copilot_directory_allowlist.py tests/launcher/test_copilot_installation.py -q→ ✅git commit(ruff, ruff format, pyright, detect-secrets, and related checks) → ✅Regressions: ✅ None detected in the targeted launcher/allowlist/install surfaces.
Step 19: Outside-In Testing Results
Test Environment: fresh isolated HOME/PATH with a stub
copilotbinary, exercised through the realuv run amplihack copilotCLI entrypointInterface Type: CLI
User Flows Tested:
uv run amplihack copilot--autopilot --yolo --max-autopilot-continues 100--autopilot,--yolo, and--max-autopilot-continues 100uv run amplihack copilot -- --max-autopilot-continues 55--autopilot/--yolodefaults and ended with--max-autopilot-continues 5Integration Points Verified: CLI passthrough parsing + launcher command assembly → ✅
Observability Check: captured argv/transcripts saved during local validation → ✅
Issues Found: initial implementation forced autopilot even when passthrough args were provided; fixed before PR creation.